The community edition is Apache-2.0 and free to self-host, permanently. Commercial tiers add enterprise governance, private-deployment licensing and support.
You pay your model provider directly. We charge by agent count, call volume, concurrency and seats — we do not resell model tokens. Whichever model you use and whatever it costs, you settle that with the provider yourself.
Four things only, and you can see your usage inside the product.
Published agents. Drafts do not count.
Total model calls, counted per calendar month.
How many runs may execute at the same time.
People who can sign in to the console.
What happens if you exceed a limit:
Unit prices and tiers are given in a formal quote. Usage inside your allowance costs nothing extra; the system warns you as you approach a limit rather than silently overcharging.
Why not charge per token? Because token prices are set by model providers and are still moving. Passing that through would make you carry a price risk that has nothing to do with us. Per-call pricing is more stable, and easier to budget for.
If you need it inside your own network and would rather not renew annually, take a perpetual licence.
A one-off purchase, priced at roughly 2.5–3× the equivalent annual subscription. The licensed major version keeps working regardless of whether you renew support.
Around 15–20% of the licence fee, covering version upgrades, issue response and security fixes. Not buying it does not disable the version you already licensed.
Billed per person-day. A standard deployment is typically 10–15 person-days, covering environment adaptation, middleware integration, agent and knowledge-base configuration, and go-live support.
We would rather work remotely than on site. On-site time is spent only on the critical points — environment review, integration, go-live, acceptance. Everything else happens remotely with documentation. That costs you less, and lets us serve more customers at once, which is better for both sides. On-site work is available when genuinely needed, but it is noticeably more expensive.
Not an industry knowledge document — a configuration asset you import.
You supply the industry facts. Refund rules, time limits and prices are only accurate when they come from you. What we deliver is the structure and the patterns, not invented business rules.
Full industry pack details, including all nine industries and the delivery steps →
Skeletons currently available:
Quarterly updates to scenarios, evaluation sets and patterns, kept in step with product releases. The same update serves every customer in that industry, which is why it can be priced far below bespoke work.
Tell us your scenario, seat count and deployment preference, and we will suggest a tier and a price. If the community edition is enough, we will say so.
Prices shown are indicative ranges; the formal quote governs. Model costs are not included and are settled directly with your model provider.